Today's wireless networks, such as HSPA (High Speed Packet Access) and LTE (Long Term Evolution), have high transmission capabilities and are robust to multipath delay that usually occurs in frequency selective fading channels. However, such performance can only be achieved where accurate channel information is available, which is seldom easily obtainable. In coherent decoding, known pilot symbols are often employed, such as in LTE for instance, where known subcarriers are inserted in the time-frequency grid. In certain communication channels the pilot density such that accurate channel estimation can be obtained. However, in other communication channels, the pilot density is so low that accurate channel estimation cannot be easily obtained. This problem can be alleviated by increasing the pilot symbol density, i.e. increasing the number of pilot symbols with respect to the number of data symbols. However, such solution may reduce the resource for the data transmission in time and/or frequency domain, thus decreasing the overall throughput.
Recently, there have been efforts to perform channel estimation by using the known pilot symbols along with unknown data symbols. Such methods, known as semi-blind, exploit the statistics of the unknown data symbols as well as the known pilot symbols in order to provide better performance than pilot based method using the same number of pilot symbols or, alternatively, requiring a smaller number of pilot symbols to achieve the same performance. Document (1) “Semi-Blind Multi-User Detection for LTE PUCCH” (Yang Hu; Astely, D.; Baldemair, R.; Falahati, S., Wireless Communications and Networking Conference, 2009. WCNC 2009. IEEE, pp. 1,5, 5-8 Apr. 2009) makes use of such method in a LTE uplink decoder, by calculating an exact maximum likelihood (ML) estimate value of the channel. However, in the foregoing LTE uplink decoder, complexity is exponential in the number of source data symbols which makes it impracticable for input information having large number of bits.
Therefore, it would be desirable to have an improved decoder that would be able to improve channel estimation accuracy without requiring high computational complexity.